On the eve of Governor Ron DeSantis‘s expected decision to begin “Phase One” of his plan to re-open the state, the Miami Herald and a public radio news outlet (WUSF) seemed dead-set against it, publishing alarmist stories that focused on the admittedly undeniable fact that Florida reported more deaths on Tuesday morning than any previous day so far. But a closer look at the underlying data reveals two mitigating facts that take the air out of the story: first, Florida changed its reporting format over the weekend, and second, over the past month, far fewer deaths get reported on weekends.
Since we know people generally don’t pick which day they die, we can surmise that some Florida jurisdictions aren’t reporting their deaths until the weekend is over. Add the two together and “Florida’s deadliest day” was likely little more than a reporting anomaly, sprinkled with just a tinge of laziness, a penchant for clicks, and maybe a dash of fear.
The Miami Herald (*sigh*) led the charge, earning a national news link from Matt Drudge, of the Drudge Report, who sent the curious to MSN.com, which has a content sharing agreement with the Herald. Drudge summed up the story in three words: “Florida’s Deadliest Day.” A screencap of the Drudge home page is presented below as Exhibit A:
The actual Miami Herald headline was only slightly better: “Death toll grows by 83, highest in 1 day.” The story even came close to touching on the likely reason for the spike, but incredibly, they focus on the opposite conclusion, deliberately choosing to highlight the negative:
Besides the high death toll, Tuesday’s number of new cases is also the highest the state has seen in the past three days. Saturday was the lowest, when the state reported a total of 306 new cases and nine deaths — a steep decline of new daily cases and deaths not seen since late March, likely caused by the Florida Department of Health changing its system of providing updates on new COVID-19 cases from twice a day to one.
So the Miami Herald looks at the same data available from the state, yet concludes the “steep decline” was “likely caused by” a change in reporting formats. Let’s take a look for ourselves (all data pulled from Florida’s Department of Emergency Management COVID-19 reporting page):
Well, well, well. What do we see here? As early as March 28th (a Saturday) and 29th (a Sunday), we can see the pattern developing. This isn’t even one of those “maybe” deals where the pattern is sorta-kinda there if you squint. It’s patently obvious. Every single weekend from that point forward, we see, for both daily hospitalizations and daily deaths, a significant drop-off in the numbers. Check it yourself: the weekend dates are obvious from the data plot, but just in case, they are: April 4th and 5th, 11th and 12th, 18th and 19th, and finally the 25th and 26th (note to Miami Herald reporters – there’s another weekend coming soon).
What is causing the dip? Do people just not want to go to the hospital on weekends? Eh…Maybe some. But the lion’s share of the drop off is likely due to the fact that most people don’t head to the hospital unless their doctor refers them. The rest – the ones that actually are reported on the weekends – probably are going to the Emergency Room. But let’s not dwell on the hospital data. The Miami Herald’s claim was focused on the drop off in deaths, which they said was likely caused by a change in reporting.
We’ll get to the change in reporting in a moment. First, let’s dig deeper into the deaths. Why is there an obvious undercount on the weekends? We don’t know for sure, but it’s likely that some hospitals and morgues in some areas simply don’t report the data – for whatever reason – on weekends. Any number of administrative reasons might be a factor. It really doesn’t matter. What does matter is the obvious conclusion we can draw from this pattern: if someone dies in Florida on the weekend from COVID-19, and their death doesn’t get reported on Saturday or Sunday, that death statistic becomes part of a backlog that will get reported later in the week.
Now the Miami Herald would read into that and likely run with a headline proclaiming “Weekend death rate much higher than previously reported,” and they’d also likely imply some sort of conspiracy by Governor DeSantis, or Florida House Speaker Jose Oliva, claiming they’re suppressing the data to paint a less scary picture.
But the trade-off is obvious. If we cram more hospitalizations and deaths into the weekdays, they’ll look worse than they should. If they get reported as-they-happen, the spikes get leveled out and we get less alarming headlines from the Herald.
Now let’s deal with the change in reporting format. For whatever reason, Florida did switch from twice-per-day reporting to just once. I wouldn’t have even minded if they would have kept the evening reports, because that’s what we’ve been relying on at The Capitolist since early March to build our spreadsheets and charts. But alas, the state switched to publishing data only once per day, and of course they chose to stick with the morning reports.
As the Herald correctly noted, on the day of the switch, which occurred on a weekend, that day’s evening report went missing. Half the day’s statistics, which include hospitalizations and deaths, got pushed into the next morning’s report. You can even see it on the chart above. The Sunday report was uncharacteristically higher than Saturday, and it’s typically not. Then came Monday, an incredibly low day, followed by a sudden spike in the data.
“But, but, but…!” some might say. “Tuesday isn’t a weekend! The spike happened on a weekday, a full two days after the weekend. This analysis is worthless!”
Calm down, I say. Tuesday’s report is published in the morning. It’s a reflection of Monday’s data, which includes all those people sent to the hospital by their doctor after staying away all weekend, and all those pour souls who succumbed to the disease over the weekend but didn’t get reported until Monday.
Yesterday wasn’t Florida’s deadliest day after all. It’s time to reopen Florida.